Modelling the contribution of biogenic VOCs to new particle formation in the Jülich plant atmosphere chamber

Introduction Conclusions References


Introduction
New particle formation, including clustering of nanoparticles and their growth to larger sizes, has been observed world-widely in continental boundary layers and free troposphere (Kulmala et al., 2004;Mirme et al., 2010) iments and model simulations indicate that gasous H 2 SO 4 plays an important role in atmospheric new particle formation, yet H 2 SO 4 alone appears not to be able to explain all the observed features of this process (Kulmala et al., 2000;Boy et al., 2003;Sipilä et al., 2010;Riipinen et al., 2007;Sihto et al., 2006;Kerminen et al., 2010;Kulmala et al., 2013Kulmala et al., , 2014)).Basic compounds like ammonia and certain amines have been proposed to act as stabilizing compounds during clustering of nanoparticles (Berndt et al., 2010;Almeida et al., 2013;Kurtén et al., 2008), while subsequent steps of atmospheric new particle formation seem to rely on the presence of low-volatile organic compounds (e.g., Metzger et al., 2010;Paasonen et al., 2010;Riipinen et al., 2012;Ehn et al., 2014;Schobesberger et al., 2013).
Oxidation products of biogenic volatile organic compounds (BVOCs) constitute the largest source of secondary organic aerosol (SOA) in the global atmosphere (Tsigaridis and Kanakidou, 2003;Hallquist et al., 2009;Spracklen et al., 2011), accounting for the main composition of SOA condensational growth (VanReken et al., 2006;Hao et al., 2009;Riipinen et al., 2012).BVOC oxidation also produces extremely low volatile organic compounds essential to the new particle formation process (Ehn et al., 2014).The most abundant group of BVOCs, accounting for more than half of their global emissions, are terpenoids (Guenther et al., 1995).Terpenoids include compounds consisting of one to several isoprene units, e.g.isoprene (C 5 H 8 ), monoterpenes (C 10 H 16 ), and sesquiterpenes (C 15 H 24 ).Oxidation products of monoterpenes have shown substantial contribution to SOA formation (Hoffmann et al., 1997(Hoffmann et al., , 1998;;Laaksonen et al., 2008), and low-volatile substances produced by sesquiterpene-ozone reactions may also initiate SOA formation (Bonn and Moortgat, 2003).Oxidation of isoprene leads to the formation of SOA (Surratt et al., 2006;Claeys et al., 2004), yet isoprene may also suppress the new particle formation process due to its high reactivity with OH (Kiendler-Scharr et al., 2009).Overall, the exact contribution of BVOCs to SOA still remains uncertain, especially with respect to the initial steps of atmospheric new particle formation.Introduction

Conclusions References
Tables Figures

Back Close
Full In this study, we used the zero-dimensional version of MALTE (Model to predict new Aerosol formation in the Lower TropospherE, Boy et al., 2006), including near-explicit air chemistry and detailed aerosol dynamics, to investigate new particle formation from a measurement campaign conducted in the Jülich Plant Atmosphere Chamber (JPAC).The in-depth analysis of the chamber measurements is discussed in another paper (Dal Maso et al., 2014).Here, we used the full chamber dataset including gas and particle phase measurements, firstly to evaluate the model, especially concentrations of gaseous compounds; secondly to test the BVOC participation mechanism in the particle formation process, identified in Dal Maso et al. (2014); and thirdly to quantify the relative contribution of ELVOCs and SVOCs of terpene oxidation products to the SOA volume production.

Measurement setup
Experiments were conducted in the JPAC located at Forschungszentrum Jülich, Germany.Detailed description regarding the chamber facility was given in previous articles (e.g., Mentel et al., 2009;Schimang et al., 2006).In addition, more details about this measurement campaign can be found from Dal Maso et al. (2014).In brief, the system consisted of two borosilicate glass chambers with teflon floors.Chambers were operated as continuously stirred tank reactors (CSTR) with teflon fans ensuring homogeneous air mixing.Each chamber housed adjustable temperature between 10 and 50 • C with a stability of ±0.5 • C. The small chamber (1150 L) served as plant chamber and was connected to the larger chamber that worked as reaction chamber (1450 L, surface-area-to-volume ratio 4.87 m −1 ).
Three small trees aging from three to four years were brought from Hyytiälä, Finland, and included Scots pine (Pinus sylvestris), Norway spruce (Picea abies), and Silver birch (Betula pendula).These trees, representing the main boreal forest species, Introduction

Conclusions References
Tables Figures

Back Close
Full were placed in the JPAC plant chamber for more than one week before the initial experiment in order to allow them to adjust to the chamber environment.Tree emission was transfered into the reaction chamber, where O 3 was added together with water vapor directly and OH was generated periodically by turning on the UV light.Discharge lamps (Osram HQI 400 W/D) were used for illumination to simulate the solar light spectrum in both chambers.Filters (OptoChem, type IR3) that reflect wavelengths between 750 and 1050 nm were used as heat shields to avoid infrared radiation inflicted plant overheating.
Ambient air was purified by an adsorption dryer (Zander, KEA 70) and a palladium catalyst ( 450• C).O 3 , NO, NO 2 and VOC levels decreased significantly after passing the purification system.The flow through the plant chamber was 115 L min −1 from which 20 L min −1 were transfered to the reaction chamber.This flow was kept nearly constant by keeping the pressure drop constant.In addition, the 10 L min −1 flow containing ozone was added, controlled by a second flow controller.Concentrations of O 3 , CO 2 and H 2 O were measured by commercial analytical instruments.Two Gas Chromatography Mass Spectrometer (GC-MS) systems were used, one to measure the VOC concentrations from C 5 to C 20 in the outflow air from the plant chamber (Heiden et al., 2003), and another to identify the OH concentration by determining the decrease in the concentration of a tracer compound in the reaction chamber (Kiendler-Scharr et al., 2009).Meanwhile, the VOC concentration was continuously measured by an on-line Proton Transfer Reaction Mass Spectrometer (PTR-MS) in the plant and reaction chamber.The gas phase H 2 SO 4 concentration in the reaction chamber was measured by a Chemical Ionization Mass Spectrometer (CIMS) (Petäjä et al., 2009;Mauldin et al., 1998).A prototype Airmodus Particle Size Magnifier (PSM) coupled with a TSI condensation particle counter (CPC) with a detection limit of ca.1.6 nm was used to count the total number concentration of particles larger than 1.6 nm ( Vanhanen et al., 2011).Also, another TSI CPC (TSI3022A) with a detection size limit of 7 nm was used to measure the total number concentration of particles larger than 7 nm.14,2014 Modelling the contribution of biogenic VOCs to new particle formation Introduction

Conclusions References
Tables Figures

Back Close
Full A Scanning Mobility Particle Sizer (SMPS TSI3071 + TSI3025A) was used to measure the particle size distribution in the size range of 14-600 nm.

MALTE-BOX model description
The MALTE model is a one-dimensional model, and its detailed description can be found in Boy et al. (2006Boy et al. ( , 2013)).In this study, we employed several modules from the original code as the MALTE-BOX model, which consists of modules for BVOC emission, air chemistry, aerosol dynamics, and wall losses.The BVOC emission module was disabled in this study instead VOC concentrations measured by PTR-MS and GC-MS were used as model input.The schematic description of the MALTE-BOX model is illustrated in Fig. 1.In the following, we will briefly introduce the main modules from the MALTE-BOX model used in our study.

Air chemistry
Chemical reaction equations were selected from the Master Chemical Mechanism (MCM) v3.2 (Jenkin et al., 1997(Jenkin et al., , 2012;;Saunders et al., 2003) via website: http: //mcm.leeds.ac.uk/MCM/.The MCM is a well-known and widely-used near-explicit chemical degradation mechanism which simplifies the chemical path of compounds by lumping products beyond the second oxidation step.We used the Kinetic Pre-Processor (KPP) version 2.1 (Damian et al., 2002) for the calculation of chemistry.Among all the compounds measured by GC-MS in the JPAC plant chamber during our experiments, 28 organic compounds were included in the air chemistry reaction calculations (see Table 1).In the table, the "other MTs" equals to the concentration difference between the summation of the concentrations of the 13 selected monoterpenes by GC-MS and the total monoterpene concentration measured by PTR-MS, while the "other SQTs" equals to the summation of other sesquiterpene isomers besides the listed four sesquiterpenes measured by GC-MS.In total, we included 2291 species and 6469 chemical and photochemical reactions for the chemical calcula-Introduction

Conclusions References
Tables Figures

Back Close
Full tions.This includes relevant inorganic reactions and the full chemistry path for isoprene, α-pinene, β-pinene, β-caryophyllene, toluene, 2-butanol and hexanal.Also, we included the first-order reactions of OH, O 3 , and NO 3 , with the following organic compounds: myrcene, sabinene, camphene, ocimene, carene, "other MTs" (which assumed the same rate coefficients as α-pinene), cineole (also known as eucalyptol), farnesene, "other SQTs" (which assumed the same rate coefficients as β-caryophyllene), α-terpinene, γ-terpinene, α-phellandrene, β-phellandrene, and terpinolene.Furthermore, we included the first-order reactions between OH and the following organic compounds: tricyclene, nonanal, and bornyl acetate.Finally, the first-order reactions between O 3 and the following organic compounds: α-humulene, α-longipinene, and ∆-cadinene were included.The full MCM chemistry path for ocimene was not available.However, we included its chemistry by approximating the chemistry beyond the first oxidation step with the chemistry path of limonene, as its structure is the closest to ocimene where the full MCM chemistry path was available.Also, we approximated the chemical path for ∆ 3carene after the first oxidation step with α-pinene, since both are bicyclic monoterpenes with an endocyclic double bond, as was done in Boy et al. (2013).H 2 SO 4 has been considered as one of the key candidates involved in the atmospheric nucleation processes.To accurately predict the H 2 SO 4 concentration using MALTE-BOX, we used the latest chemical reaction rates for Criegee Intermediates (CI) reactions in the air chemistry module, which included CI chemistry for α-pinene, βpinene, and limonene as in scenario C in Boy et al. (2013) and the CI chemistry for isoprene as in scenario D in Boy et al. (2013).

Aerosol dynamics
Aerosol dynamic processes, including nucleation, condensation, evaporation and coagulation, were calculated using the size-segregated UHMA model (University of Helsinki Multicomponent Aerosol model, Korhonen et al., 2004), which has been implemented in the MALTE-BOX model as a module.We used the kinetic H 2 SO 4 nucleation Introduction

Conclusions References
Tables Figures

Back Close
Full mechanism from the original UHMA code, in which the particle formation rate (J) is described as a power-law with the H 2 SO 4 concentration (Kulmala et al., 2006): where K is termed as "kinetic nucleation coefficient", including both the collision frequency and the probability of formation of a stable particle after the collision (Paasonen et al., 2010).In addition, based on the findings in Dal Maso et al. ( 2014), we updated the UHMA code to include the nucleation mechanism involving oxidation products of BVOCs, and we refer to it as organic nucleation in this study.As described by Dal Maso et al. ( 2014), data analysis showed that assuming both H 2 SO 4 and oxidation products of BVOCs are required for formation and activation of clusters produced corresponded best with observed results.Such a mechanism is described by the equation: where P is the nucleation coefficient for organic nucleation (Paasonen et al., 2010).
[ELVOC] refers to the concentration of ELVOC.In this study, we set the critical nucleation cluster size to be 1.5 nm in all the model runs (Kulmala et al., 2007;Zhang, 2010;Wang et al., 2013).Schobesberger et al. (2013) found that both the H 2 SO 4 and oxidation products of monoterpenes were involved in the nucleation process.Ehn et al. (2014) suggested that ELVOC from the oxidation of α-pinene participated in the initial steps of new particle formation.However, both studies were from well-controlled chamber experiments, with only a single BVOC compound investigated.Therefore, the identification of the ELVOC in the nucleation process remains uncertain, especially from mixtures of biogenic emissions.In this study, we did not try to identify any specific compounds involved in the nucleation, but rather aimed at finding differences in aerosol formation rates after the introduction of VOCs, as the compounds from MCM are too volatile to be in the range of ELVOC.Thus, we used representative compound ELVOC OH (ELVOC from OH oxidations) as input of [ELVOC].Introduction

Conclusions References
Tables Figures

Back Close
Full The information on e.g., saturation vapour pressures of condensable gas compounds is still insufficient and under investigation (Topping et al., 2007).Previous studies from the JPAC chamber experiments suggest that large amounts of ELVOCs is being formed from VOC oxidation, and that the reaction products contribute to the particle growth (Ehn et al., 2014).Here we used a simplified gas-particle partitioning model which classifies the terpene oxidation products into either ELVOCs or SVOCs.The reaction mechanisms can be described with following equations, (3) We then selected several compounds as the representatives of ELVOCs and SVOCs.Several criteria were used to select these compounds.First, compounds contained at least 9 carbon atoms, similar to the finding by Schobesberger et al. (2013).Second, selected compounds had the highest concentration among the VOC oxidation products, which should relatively represent the distribution of their reactants.We selected two compounds from the first stable oxidation products of reactions between VOCs and OH where the full chemistry path was known, and then we defined these two compounds as the representatives of the ELVOC OH and SVOC OH .As a result, ELVOC OH was APINAO2 (MCM name, oxidation product of α-pinene by OH) and SVOC OH was BPINAO2 (MCM name, OH oxidation product of β-pinene) in this study.The same approach was applied to the first stable oxidation products of reactions between VOCs and O 3 .Pinonic acid was then selected as ELVOC O 3 and nopinone was chosen as SVOC O 3 .From the first stable oxidation products of reactions between VOCs and NO 3 , we chose only NAPINAO2 (MCM name, NO 3 oxidation product of α-pinene) as SVOC NO 3 , and no ELVOC was taken into account.
To run the model, we set the saturation concentrations of ELVOC OH and ELVOC O 3 to be three times larger than the H 2 SO 4 vapour (1 cm −3 ), the same value as in Boy et al.Introduction

Conclusions References
Tables Figures

Back Close
Full  2006), and the saturation concentrations of SVOCs to be 5 × 10 7 cm −3 .In order to not systematically underestimate the total particle volume concentration, the formation of the five condensable compounds were multiplied with a fixed correction factor of two.We can then compare the modelled aerosol volume with the measurements, in a similar manner as Wang et al. (2013).
We used the fixed sectional approach to represent the particle size distribution in our simulation (Korhonen et al., 2004), which included 40 particle size bins.

Chamber wall loss
The original MALTE-BOX model includes two parameterizations for particle wall losses: one was from Crump and Seinfeld (1981) for mono-dispersed particles, and another was poly-dispersed particle theory by Park et al. (2001).We used the poly-dispersed particle loss mechanism in this study.The first order gas-wall deposition k g, w (s −1 ) was also calculated from Park et al. (2001).The reversible gas-wall partitioning of SVOCs was modelled by the following equation, where k w, g (s −1 ) is the desorption rate of SVOCs from the chamber surfaces out to the gas phase, C w is an effective wall equivalent mass concentration of solvent which the organic compounds can dissolve into, M w is the average molar mass of the solvent on the walls, and γ w,i is the activity coefficient of compound i in the solvent.Compared We assumed that ELVOCs should completely stick on the chamber surfaces, thus such reversible process was not considered for ELVOCs in this study.

Model input parameters
We executed the MALTE-BOX model for one measurement campaign conducted in the JPAC chamber.For analysing model results, we chose the first four continuous days from the whole measurement campaign because these days had continuous measurements, coverage of the dataset was complete, UV light-on period was the same, and chamber was stable.
The measured temperature and relative humidity (RH) in the JPAC reaction chamber were used directly in the model as input.As an example, values from the first four days are illustrated in Fig. 2. The temperature stabilized around noon on Day-1, after that the temperature was ca.16 • C for all days.The RH had minor fluctuations during the experiment.RH was kept to ca. 60 % during UV-on periods for the first three days.On Day-4, the RH was around 55 %.In addition, 8 discharge lamps were used to simulate solar illumination on Day-1, but only 4 lamps were used on the remaining days.
Because ambient air in the JPAC chamber was purified by an adsorption dryer, the NO 2 concentration was set to 200 ppt v , and the NO concentration was set to 30 ppt v (Schimang et al., 2006).Other inorganic gases such as CO and SO 2 were not measured, thus some estimated constant values were used as input in the model.All the inorganic compound concentrations in the inflow used in the MALTE-BOX model are listed in Table 2. Concentrations of VOCs were based on the data from GC-MS and PTR-MS, and concentrations of VOC oxidation products were set to zero initially.As an example, concentrations of isoprene, monoterpenes and sesquiterpenes measured from the first four days are illustrated in Fig. 3.The monoterpene and sesquiterpene concentrations displayed a certain diurnal distribution pattern, because the discharge lamps mimicked sun light in the plant chamber.The total monoterpene concentration exceeded 4 ppb v on Day-1, and then decreased to a value below 2 ppb v on the following three days.In particular, the ocimene concentration was the highest on Day-1, which Introduction

Conclusions References
Tables Figures

Back Close
Full can be explained by the fact that ocimene emission is light dependent (Owen et al., 2002), and the simulated solar light intensity from the discharge lamps was doubled on Day-1 in the plant chamber.
The isoprene concentration and the total sesquiterpene concentration were about an order of magnitude lower than the monoterpene concentration.Overall, the measured concentrations of terpenes were at similar levels as ambient air concentrations in boreal forest environments (Ruuskanen et al., 2009).
For particle data input, only the first measured particle size distribution from SMPS measurements was used for initialization of the model.In addition, the condensation sink calculated from SMPS measurements was used in the model to account for the condensation loss of sulphuric and nitric acid (Boy et al., 2006).

O 3 and OH concentrations
In the JPAC reaction chamber, high OH concentrations were generated by periodically switching on the UV light.The UV light intensity in the JPAC reaction chamber was assumed stable and was set constant in our first model runs.The modelled O 3 concentration corresponded well with the measurements when the UV light was off (Fig. 4a).The modelled O 3 concentration reached steady state at the same level as the measurements, indicating that the dark reactions of VOCs were well represented in the model.Although the measured O 3 concentration showed a decrease at the beginning of Day-1, both modelled and measured O 3 concentrations reached levels of ca.57 ppb v before the UV light was switched on.The O 3 concentration decreased sharply to about half of its initial value after the UV light was turned on, and then a continuous decrease in O 3 concentration was observed before the UV light was turned off.The modelled O 3 reached a stable plateau after the sharp decrease (see the dash green line), whereas the measured O 3 concentration continued to decrease until the UV light was switched off.As an explanation, the oxidation products from previous experiments may have been deposited and oligomerized on the lamp or chamber walls, and then photolyzed or thermally cracked introducing firstly a small time dependent extra sink for O 3 in the chamber, and secondly a small increase in UV light intensity (this phenomenon disappeared after cleaning the UV lamp and did not reappear in succeeding campaigns, since the UV light lamp was placed inside a quartz tube with no direct contact to the reactants in the reaction chamber).To simulate this extra sink, we increased the UV intensity linearly in time so that the observed O 3 was matched during phases with UV turned on.By applying 80 % UV light intensity increase, the modelled O 3 concentrations (red dots in Fig. 4a) displayed the same distribution as the measured O 3 concentration during UV-on periods.The measured O 3 concentration decreased slightly more from Day-1 to Day-4.Nevertheless, we could reproduce the observations by applying the same 80 % UV light intensity increase for all four days.
The time series of modelled OH concentration is illustrated together with the estimated OH concentration by GC-MS in Fig. 4b on Day-3 and Day-4.The values scattered around ±50 %, which was well within the estimated error of a factor of two for the marker method.The modelled OH concentration level was similar to the measured one, being on average about 7 × 10 7 cm −3 .The modelled OH concentration was lower on Day-1 compared with the rest of the period.This is consistent with the highest BVOC concentrations measured on Day-1, leading to a higher consumption of OH with BVOC reactions.Overall, the OH concentration in the JPAC reaction chamber was an order of magnitude higher than the measured ambient OH concentration e.g., in the boreal forest environment (Petäjä et al., 2009).

H 2 SO 4 concentration
Here we present the modelled H 2 SO 4 concentration, as H 2 SO 4 is needed for homogeneous nucleation.The time series of the modelled H 2 SO 4 concentration is depicted in Fig. 5, together with CIMS measurements.The measured H 2 SO 4 concentration was in the range 2-3 × 10 5 cm −3 during the UV-off periods.The H 2 SO 4 concentration was slightly underestimated by the model, except during a several-hour interval on Day-1.
Once OH was generated, the modelled H 2 SO 4 concentration first followed the measured H 2 SO 4 concentration closely.H 2 SO 4 was quickly formed after the UV light was turned on, and a strong peak formation of H 2 SO 4 was observed, although our model predicted 20-30 % higher concentrations than measured.After the initial peak, however, the measured H 2 SO 4 concentration increased and the model did not capture this increase, remaining almost stable.Boy et al. (2013) have shown that reactions of stabilized CI with SO 2 could contribute to increased H 2 SO 4 yields in the range of 33 to 46 %.We included the up-to-date CI reaction rates, yet we could not fully predict the temporal behavior of the H 2 SO 4 concentration as observed.One reason to explain such a difference could be an increase in the SO 2 source rate, as we did not have the SO 2 measurements and used a constant SO 2 concentration input.Also, uncertainties related to CIMS measurement should be considered.Regardless of the second measured H 2 SO 4 concentration increase, the average H 2 SO 4 concentration after the first intensive peak showed a tendency to increase from 27986 Introduction

Conclusions References
Tables Figures

Back Close
Full Day-1 to Day-4, captured by both the measurements and the modelled results.This was probably due to the increase in the OH concentration, thus leading to greater SO 2 oxidation.Because of relatively large differences between the model and measurements, we used the measured distribution of H 2 SO 4 concentration as model input for the following particle phase simulations to diminish the uncertainty from the gas phase simulations, as H 2 SO 4 was among the critical parameters in the nucleation process in the MALTE-BOX model.

Time series of BVOC concentrations
To investigate the potential contribution of BVOCs to the SOA formation in this chamber study, it is essential to properly predict the time series of the BVOC concentration.By compiling the most up-to-date chemical reaction rates for BVOCs in the MALTE-BOX model, the resulting concentrations of isoprene, monoterpenes, and sesquiterpenes modelled in the JPAC reaction chamber are plotted in Fig. 6, together with isoprene and total monoterpene concentrations measured by the PTR-MS.The modelled isoprene concentration showed a similar distribution as the measurements during the UV-off period on Day-1.The isoprene concentration in the model decreased to below 0.01 ppb v when the UV light was switched on.The decrease in the measured isoprene concentration was not as pronounced as in the modelled results.Also, because of some missing input data points at the beginning of these three continuous days (seen in Fig. 3), the modelled isoprene concentrations were significantly lower than the measured concentration.
The total monoterpene concentration measured by the PTR-MS is illustrated with blue markers in the middle panel in Fig. 6.The summation of the modelled monoterpene isomer concentrations reached the same level as the measurements after the initial model stabilization at the beginning.The monoterpene concentrations in the reaction chamber decreased to about one third of their concentrations measured in the plant chamber because of the in-flow dilution and chemical reactions with O 3 .When the UV light was switched on, the monoterpene concentration decreased sharply due 27987 Introduction

Conclusions References
Tables Figures

Back Close
Full to OH oxidation, as seen both from the measurements and simulation results in Fig. 6.The modelled monoterpene concentration showed a more pronounced decrease during the UV-on periods than measurement.One possible reason for such an overestimation of the monoterpene reactions could be that the reaction coefficient for the other MTs' was overestimated to some extent.Overall, the modelled monoterpene concentrations corresponded well with the measurements, so it is reasonable to use the modelled monoterpene concentrations and their oxidation products for the particle phase simulation.
The sesquiterpene concentrations were beyond the detection capability of PTR-MS in this study.Our modelled results stated that sesquiterpenes were strongly oxidized by O 3 once the tree emission from the plant chamber (Fig. 3) entered the reaction chamber.The sesquiterpene concentrations decreased rapidly to below 0.01 ppb v .In addition, the sesquiterpene concentrations showed minor decrease during UV-on periods because of the oxidation by OH.Although sesquiterpenes were strongly oxidized from the loading moment in the JPAC reaction chamber, the new particle formation event only occurred during UV-on periods.Thus, we assumed that sesquiterpenes and their ozone-oxidation products were not sufficient to solely trigger the new particle formation event.

Simulation on new particle formation
We focused on two nucleation mechanisms in this study: kinetic H 2 SO 4 nucleation and organic nucleation.We used the "best-fit" approach to define the nucleation coefficients in the MALTE-BOX model, by leveling the modelled total number concentration of particles with the measurements on Day-1, and then we used the same nucleation coefficients to predict the aerosol distributions on the following days.The determined nucleation coefficient by the kinetic H 2 SO 4 nucleation was 2.1 × 10 −11 cm 3 s −1 in the model, which was an order of magnitude larger than the average value 1.4 × 10 −12 cm 3 s −1 from Paasonen et al. ( 2010), yet smaller than the maximum K value reported by Riipinen et al. (2007).The organic nucleation mechanism showed better agreement with Introduction

Conclusions References
Tables Figures

Sub-3 nm particle formation
The time series of the total number concentrations of particles larger than 1.6 nm (D p > 1.6 nm) measured by PSM and modelled by both nucleation mechanisms are illustrated in Fig. 7. Two particle number concentration peaks were distinguished by the PSM measurements on the first three days when the UV light was turned on: one intensive particle burst followed by another moderate peak of formed particle numbers.Only a gradual increase in the particle number concentrations was found on Day-4.The particle number concentrations modelled by the kinetic H 2 SO 4 nucleation showed an opposite behaviour, increasing continuously during the UV-on periods on each day.The modelled total particle number concentrations from the organic nucleation mechanism showed similar distributions as the measurements on Day-1 and Day-4, with ±20 % difference.The model did not predict properly the first intensive peak of particle number concentrations on Day-2 and Day-3.
Figure 7 shows that the maximum of particle number concentrations modelled by the organic nucleation mechanism gradually decreased from Day-1 to Day-4, similar to the PSM measurements, whereas an opposite relation was seen from the average particle number concentrations modelled by the kinetic H 2 SO 4 nucleation.This agrees with the previous finding from Riccobono et al. (2014) that H 2 SO 4 and ELVOCs together play a major role in the new particle formation event.However, we did not see a better agreement between the modelled results and measurements when using the proposed nucleation mechanism (J = k m [H 2 SO 4 ] 2 [ELVOC]) by Riccobono et al. (2014).This is possibly due to the remaining difficulty of identifying which specific ELVOC compounds will contribute to the new particle formation.
The scatter plots of the total sub-3 nm particle number concentrations between model simulations and measurement are illustrated in Fig. 8, and the coefficients of Introduction

Conclusions References
Tables Figures

Back Close
Full on Day-1 from the kinetic H 2 SO 4 nucleation and 0.77 on Day-3 from the organic nucleation.Overall, the determination coefficients from the organic nucleation showed relatively larger R 2 than the coefficients from the kinetic H 2 SO 4 nucleation.
Although the modelled results could not fully represent the distributions of sub-3 nm particle number concentrations with any of the nucleation mechanisms, the simulations indicate that H 2 SO 4 and ELVOC OH were involved in the formation and growth of the homogeneously nucleated particles.Some other compounds, e.g.amines, may have also contributed to this process (Almeida et al., 2013;Kurtén et al., 2008), but the current model did not apply such chemistry, due to the lack of the amine measurements during this campaign.

SVOC gas phase-chamber wall partitioning and particle size distribution
In order to capture the condensational growth of formed particles, it is important to properly model the concentrations of gaseous precursors.By coupling the gas-wallpartitioning mechanism of SVOCs, we were able to model the total adsorption and desorption of SVOCs on the JPAC reaction chamber walls during the whole period.The total amounts of SVOCs adsorbed on the chamber walls, including SVOC OH and SVOC O 3 , are illustrated in Fig. 9.The SVOCs desorbed from the chamber walls are also depicted in the same figure together with the concentrations of the total VOC oxidation products by OH and O 3 .The SVOCs adsorbed on the chamber walls increased significantly during the whole period, and their concentrations reached over an order of magnitude higher on Day-4.The increase in the SVOC O 3 concentration was stronger than in the SVOC OH concentration.
The desorption of SVOCs from the chamber walls became substantial after Day-1, and reached a similar level as the summation of the ELVOCs and SVOCs which had not been deposited on the chamber walls.Introduction

Conclusions References
Tables Figures

Back Close
Full The resulting contour plots of the time evolution of modelled particle number size distributions are depicted together with SMPS measurements in Fig. 10 for the kinetic H 2 SO 4 nucleation, and in Fig. 11 for the organic nucleation.For comparison, we executed also the model without gas-wall-partitioning of SVOCs, and the corresponding results are illustrated in Figs.10b and 11b.The modelled time evolution of the particle size distributions from both nucleation mechanisms showed similar distributions as the measurements during the UV-on periods.However, the modelled particle size distributions from the organic nucleation mechanism showed better agreement with the measurements during the UV-off periods, as the particle number concentration was strongly over predicted by the kinetic H 2 SO 4 nucleation mechanism.Comparing with the modelled particle number size distributions without tracking the SVOC gaswall-partitioning, it is conclusive that a substantial amount of SVOCs adsorbed on the JPAC reaction chamber walls during the measurement campaign, and the desorption of SVOCs from the walls significantly contributed to the growth of the freshly formed particles.This finding is in line with the results from Roldin et al. (2014); Zhang et al. (2014); Kokkola et al. (2014) in that the SVOC losses on chamber wall were reversible.Nevertheless, the model assumptions and the quantitative extent of the wall effects still need experimental verification e.g. by wall adsorption/desorption studies in JPAC.In any case it is clearly important to include such adsorption/desorption mechanism to properly describe the actual SOA formation from chamber measurements.

Particle volume production and compositions
We calculated the total particle volume concentrations from the SMPS measurements and modelled particle size distributions by assuming particles are spherical.The resulting time series of the particle volume concentration is illustrated in Fig. 12.Also, the scatter plots of the total particle volume concentrations between the modelled results from both nucleation mechanisms and the measurements are depicted in Fig. 13 for a closer look.Although the model underestimated the total particle volume concentration on Day-2, we were able to reproduce the aerosol volume production similar to the 27991 Introduction

Conclusions References
Tables Figures

Back Close
Full measurements using both nucleation mechanisms overall, by considering a reversible SVOC gas-wall partitioning.
To investigate the average composition of particles during the different days, we took the total volume composition of particle volumes from average size distribution during the UV-on periods, and summed the compositions into three groups including H 2 SO 4 , ELVOCs, and SVOCs.The corresponding volume fractions of those three groups are depicted in Fig. 14.The x axis includes the first 30 particle size bins from the modelled particle size distribution, equalling particle size ranges from 1.5-250 nm in each subfigure.H 2 SO 4 accounted for 100 % in the first particle size bin in all sub-figures.This was due to the technical solution of the MALTE-BOX model, as the initial activation clusters have been set as H 2 SO 4 .However, the volume fraction of H 2 SO 4 became totally negligible beside the first particle size bin, because the H 2 SO 4 concentration was several orders of magnitude lower than the ELVOC concentrations during the simulations, and we set the vapour pressure of H 2 SO 4 and ELVOCs to a similar level.
The topmost two panels show that ELVOCs dominated the total volume composition for small particles, while contributing around half of the particle volume compositions at larger particle sizes on Day-1 from both nucleation mechanisms, similar to the previous finding by Ehn et al. (2014).However, the average volume fraction of SVOCs to the total particle volumes changed from Day-1 to Day-4, with increasing amount of SVOCs desorbing from the chamber walls.

Summary and conclusion
In this study, we used the MALTE-BOX model to extend the earlier investigation on the formation mechanisms of aerosol particles from one measurement campaign conducted in the Jülich Plant Atmosphere Chamber.For the first time, we modelled the chemical and physical processes integrating both gas and particle phase simulations from the JPAC chamber measurements.Introduction

Conclusions References
Tables Figures

Back Close
Full The gas phase simulation results showed that the modelled and measured H 2 SO 4 concentrations agreed within a factor of two.The modelled OH concentration was within the ±50 % uncertainty range of the estimated OH concentration from a tracer compound using GC-MS.The modelled O 3 concentration was within 10-20 % from the measurements.Monoterpene concentrations predicted from the model agreed well with the total monoterpene concentrations detected by PTR-MS.In addition, the model has shown that sesquiterpenes were strongly oxidized by O 3 in the JPAC reaction chamber and minor oxidations by OH, even though the PTR-MS was not capable of providing such information on sesquiterpenes.
Our results suggest that H 2 SO 4 is one of the key compounds involved in the new particle formation, but cannot solely explain the new particle formation process.The nucleation coefficient predicted from the kinetic H 2 SO 4 nucleation was over an order of magnitude higher than in previous studies, while the organic nucleation coefficient better explained the experimental data, and its values were comparable with other studies.
We found that the SVOCs adsorbed on the chamber walls partly desorbed and substantially contributed to the SOA mass formation during Day-2 to Day-4 of the experiment.The modelled results, including particle growth and particle volume, strongly indicate that with every additional day of the model runs, all processes in the chamber were influenced by such chamber wall effects, which is in line with current discussions about wall effects in simulation chambers.As the SVOC contributes to SOA formation only at larger particle sizes, and the organic nucleation mechanisms depend on the ELVOC, results concerning new particle number formation are unaffected by the SVOC gas-wall-partitioning.
Overall, we have simulated the near-ambient atmospheric environment with the most up-to-date air chemistry and aerosol dynamics, the actual tree-emitted compounds were represented with model compounds, so our chemical reactions may not fully represent the actual chemistry in the JPAC reaction chamber.Our model was capable of predicting the average concentrations of both gas and particle phase components, Introduction

Conclusions References
Tables Figures

Back Close
Full being within a factor of 1-2 from measurements, yet the whole distribution of concentrations did not show a full agreement.Therefore, in order to better understand the processes in the JPAC chamber, it is important to include explicit air chemistry calculations.Also, we had simplified the calculations of particle condensational growth using representative compounds from BVOC oxidation products to grow the formed nucleation clusters in our current simulations.To better recognise the actual growth of aerosol particles via condensation, it is necessary to include all condensable vapour precursors with their saturation vapour pressure information in future studies.Full Field observations, laboratory exper-Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | with the values determined byMatsunaga and Ziemann (2010) for teflon chambers, we needed to use a ca.3-4 order of magnitude smaller value (1 × 10 −8 mol m −3 ) for the term C w /(M w γ w,i ) in order to capture the observed reversible wall losses in the JPAC glass reaction chamber.Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | . During UV-off periods, OH concentration was in the range 10 5 -10 6 cm −3 , while after the UV light was turned on, OH concentration rapidly increased.The modelled OH concentration showed similar values as the estimated OH concentration after the first one-hour when the new steady state was well established.The transient of the OH concentration in the beginning cannot be detected by the marker measurements, because of limited time resolution of the GC-MS instrument.The second moderate increase in the OH concentration predicted by the model agreed better with the measurements, by applying 80 % UV light intensity increase during UV-on periods.In contrast, a decreasing trend in the OH concentration during UV-on periods was predicted by the model with constant UV light intensity.This justifies our previous assumption on increasing the UV light intensity.The experimentally determined OH concentration, after an hour when the UV light had been turned on, was 5 × 10 7 cm −3 on Day-1, about 6 × 10 7 cm −3 on Day-2, and about 7-8 × 10 7 cm Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | listed in the figure legends.The smallest R 2 value was 0.7 Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Table 1 .
The VOCs measured by GC-MS which were included as input in the MALTE-BOX model for the air chemistry module.The "other MTs" and "other SQTs" refer to other monoterpenes and sesquiterpenes than those specified in the table, respectively.
L. Liao et al.

Table 2 .
The constant concentration and flow rate of each inorganic gas compound as input in the MALTE-BOX model.